SOCW 671 #11 Correlation and Regression
Uses of Correlation To study the strength of a relationship To study the direction of a relationship Scattergrams are used to display data
Types of Correlation Perfect Correlations: -1.0 or 1.0 Nonperfect Correlations: everything else Strength and Direction: - or +; near 0.0 weak; near –1.0 or 1.0 strong Represented by “r.” Interpreting Linear Correlations Correlation Continuum: -1.0 to 1.0
Coefficient of Determination Proportion of the variation in one variable that accounts for (or related to) the variation in the other variable R-squared).
Using Correlation for Description Can be used as a mathematical alternative to the scatterplot. Pearson Product-Moment Coefficient: a parametric measure of the strength of a correlation. Need two interval or ratio level variables. Spearman’s rho: a nonparametric measure that uses ordinal level variables
Bivariate & Multiple Correlation Bivariate (two variables) vs. Multiple Correlation (three or more variables) Partial R: removes the influence of a third (intervening) variable Multiple R: uses one criterion (dependent) variable and two or more predictor (independent) variables.
Regression Can be used to predict explain describe
Regression Line Computation of the regression line: Y’ = a + b(X) Regression Coefficient (b): slope of the line The y-Intercept (a): the starting point of the line Predicted Y (Y’): estimated score for the criterion value
Least-squares Criterion The process of fitting a line to the data Minimizes the differences of expected to actual values
Selecting Correct Statistical Test Five considerations Sampling methods Distribution of the dependent (and sometimes the independent variable) within the population Level of measurement of the independent and dependent variables Statistical power of the test Robustness of the test